Skip to main content

​
What is Model Context Protocol (MCP)?

MCP is an open protocol that standardizes how applications provide context to LLMs. Think of MCP like a USB-C port for AI applications. Just as USB-C provides a standardized way to connect your devices to various peripherals and accessories, MCP provides a standardized way to connect AI models to different data sources and tools. Read more here.

​
Setting Up Your Workspace MCP server

When you set up a flow, you can use tools beginning with INPUT_ to set the inputs that your tool will have. For example, 
- id: word_1
  tool: INPUT_TEXT
  input:
    - name: value
      value: DEFAULT_VALUE
    - name: data_type
      value: STRING
- id: word_2
  tool: INPUT_TEXT
  input:
    - name: value
      value: DEFAULT_VALUE
    - name: data_type
      value: STRING
- id: concat
  tool: PYTHON_SANDBOX_RUN
  needs:
    - word_1
    - word_2
  input:
    - name: code
      value: |

        word_1 = steps.word_1.result.output
        word_2 = steps.word_2.result.output

        f"{word_1} {word_2}"
    - name: data_type
      value: STRING
The above flow would show the parameters word_1 and word_2 for the MCP tool. Please note that tool IDs, not names, are used to reference the tools in the flow. After you add these INPUT_ tools, you can navigate to the MCP tab of your workspace to configure flows to be used for MCP, set prompts for the flows to describe what they do, and set the prompts for the parameters of your configured flows as well.

​
Connecting to LLM

To connect your workspace MCP to an LLM, use the sample configuration found on the MCP tab of your workspace. Remember to replace the secret key! If you have no other MCP servers, you can paste the whole file, otherwise, you can add just the jinbaflow section (for Jinba Flow) to your file. Here is an example of a configuration file: 
{
  "mcpServers": {
    "jinbaflow": {
      "command": "npx",
      "args": [
        "-y",
        "supergateway",
        "--sse",
        "https://api.jinba.io/api/v1/mcp/uuid-here-1234-5678-foobar/sse",
        "--header",
        "Authorization: Bearer YOUR_TOKEN"
      ]
    }
  }
}